[Skip to Navigation]
Sign In
Figure 1.  Between-Class Analysis of the Association Between Selective Serotonin Reuptake Inhibitors (SSRIs) and Type 2 Diabetes (T2D)
Between-Class Analysis of the Association Between Selective Serotonin Reuptake Inhibitors (SSRIs) and Type 2 Diabetes (T2D)

All estimates reflect the hazard of T2D among children and adolescents treated with SSRIs compared with those without antidepressant treatment. In the intention-to-treat (ITT) analysis, the graph shows results of the unadjusted analysis, the analysis partially adjusted for demographics (age, sex, and race for the Medicaid Analytic eXtract database only) and treatment indication, and the analysis fully adjusted for all baseline covariates. In the as-treated analysis, the graph shows the results of the unadjusted analysis, the analysis adjusted for all baseline covariates, and the analysis adjusted for all baseline and time-varying covariates. For subgroups, all estimates reflect the hazard of T2D among children and adolescents treated with SSRIs compared with those without antidepressant treatment, fully adjusted for all baseline covariates. Propensity score models were reestimated within each subgroup. HR indicates hazard ratio.

Figure 2.  Weighted Survival Curves for Probability of Remaining Free of Type 2 Diabetes
Weighted Survival Curves for Probability of Remaining Free of Type 2 Diabetes

In intention-to-treat analyses, Kaplan-Meier curves were weighted by the inverse of the propensity score (estimated probability of initiation of selective serotonin reuptake inhibitor [SSRI] treatment conditional on baseline covariates). In as-treated analyses, survival probabilities were weighted by stabilized treatment and censoring weights based on baseline and time-varying covariates and standardized to the joint distribution of baseline covariates.

Figure 3.  Within-Class Analysis of the Association Between Medications in the Selective Serotonin Reuptake Inhibitor (SSRI) Class and Type 2 Diabetes
Within-Class Analysis of the Association Between Medications in the Selective Serotonin Reuptake Inhibitor (SSRI) Class and Type 2 Diabetes

Intention-to-treat analysis compared each individual SSRI medication vs fluoxetine hydrochloride. The arrow indicates that the upper end or the lower end of the 95% CI is beyond the scale (0.5 to 2).

Table 1.  Baseline Characteristics of Patients With SSRI Treatment and Without Antidepressant Treatment in the Intention-to-Treat Populationa
Baseline Characteristics of Patients With SSRI Treatment and Without Antidepressant Treatment in the Intention-to-Treat Populationa
Table 2.  Unadjusted Absolute Rate of Type 2 Diabetes Among Children and Adolescents With SSRI Treatment and Without Antidepressant Treatment
Unadjusted Absolute Rate of Type 2 Diabetes Among Children and Adolescents With SSRI Treatment and Without Antidepressant Treatment
Supplement.

eTable 1. Conditions for Cohort Inclusion and Exclusion

eTable 2. Study Definitions for Exposure and Comparator Groups

eTable 3. Study Definitions for Exposure and Comparator Groups

eTable 4. Baseline Characteristics of Children and Adolescents With SSRI Exposure Compared to Those Without Antidepressant Exposure in the Medicaid Analytic eXtract (MAX) ITT Population

eTable 5. Baseline Characteristics of Children and Adolescents With SSRI Exposure Compared to Those Without Antidepressant Exposure in the IBM MarketScan ITT Population

eTable 6. Baseline Characteristics of Children and Adolescents With SSRI Compared With Bupropion Exposure in the Medicaid Analytic eXtract (MAX) ITT Population

eTable 7. Baseline Characteristics of Children and Adolescents With SSRI Compared With Bupropion Exposure in the IBM MarketScan ITT Population

eTable 8. Baseline Characteristics of Children and Adolescents With SSRI Compared With Psychotherapy Exposure in the Medicaid Analytic eXtract (MAX) ITT Population

eTable 9. Baseline Characteristics of Children and Adolescents With SSRI Compared With Psychotherapy Exposure in the IBM MarketScan ITT Population

eTable 10. Unadjusted Absolute Rate of Type 2 Diabetes Among Children and Adolescents With SSRI Exposure or in a Comparator Group (Untreated, Bupropion, Psychotherapy)

eTable 11. Hazard Ratios and 95% Confidence Intervals Comparing the Incidence of Type 2 Diabetes Between SSRI-Exposed and Untreated Children and Adolescents: Different Levels of Adjustment

eTable 12. Adjusted Risk Differences Comparing the Incidence of Type 2 Diabetes Between SSRI-Exposed and Untreated Children and Adolescents at 2 and 5 Years of Follow-up

eTable 13. Hazard Ratios and 95% CIs Comparing the Incidence of Type 2 Diabetes Between SSRI- and Bupropion-Exposed Children and Adolescents: Different Levels of Adjustment

eTable 14. Hazard Ratios and 95% CIs Comparing the Incidence of Type 2 Diabetes Between SSRI- and Psychotherapy-Exposed Children and Adolescents: Different Levels of Adjustment

eTable 15. Hazard Ratios and 95% CIs Comparing the Incidence of Type 2 Diabetes Within the SSRI Class vs Fluoxetine

eTable 16. Hazard Ratios and 95% CIs Comparing the Incidence of Type 2 Diabetes Between SSRI-Exposed and Untreated Children and Adolescents: Subgroup Analyses in ITT Population

eTable 17. Hazard Ratios and 95% CIs Comparing the Incidence of Type 2 Diabetes Between SSRI-Exposed and Untreated Children and Adolescents: Sensitivity Analyses in ITT Population

eTable 18. Hazard Ratios and 95% CIs Comparing the Incidence of Type 2 Diabetes Between SSRI-Exposed and Untreated Children and Adolescents: Sensitivity Analyses in As-Treated Population

eTable 19 Flowchart of the Claims Electronic Health Records–Linked Study Population (Medicaid Analytic eXtract Claims Linked to the Partners Healthcare Research Patient Data Registry)

eTable 20. Patient Characteristics and Body Mass Index Data for Children and Adolescents With SSRI Exposure and Without Antidepressant Exposure in the Claims Electronic Health Records (MAX-RPDR)–Linked Subset

eTable 21. Patient Characteristics and Body Mass Index Data for Children and Adolescents With SSRI and Bupropion Exposure in the Claims Electronic Health Records (MAX-RPDR)–Linked Subset

eTable 22. Patient Characteristics and Body Mass Index Data for Children and Adolescents With SSRI and Psychotherapy Exposure in the Claims Electronic Health Records (MAX-RPDR)–Linked Subset

eFigure 1. Study Design Diagrams for the ITT and As-Treated Analyses

eFigure 2. Potential Role of Obesity as an Unmeasured Confounder on the Observed Hazard Ratios and 95% CIs in the ITT Population

eMethods. Quantification of the Potential Impact of Residual Confounding by Poorly Measured Obesity

eReferences

1.
Dwyer  JB, Bloch  MH.  Antidepressants for pediatric patients.   Curr Psychiatr. 2019;18(9):26-42F.PubMedGoogle Scholar
2.
Mills  JA, Strawn  JR.  Antidepressant tolerability in pediatric anxiety and obsessive-compulsive disorders: a Bayesian hierarchical modeling meta-analysis.   J Am Acad Child Adolesc Psychiatry. 2019;S0890-8567(19)32113-6. doi:10.1016/j.jaac.2019.10.013 PubMedGoogle Scholar
3.
Dobson  ET, Strawn  JR.  Pharmacotherapy for pediatric generalized anxiety disorder: a systematic evaluation of efficacy, safety and tolerability.   Paediatr Drugs. 2016;18(1):45-53. doi:10.1007/s40272-015-0153-1 PubMedGoogle ScholarCrossref
4.
Morkem  R, Williamson  T, Patten  S,  et al.  Trends in antidepressant prescribing to children and adolescents in Canadian primary care: a time-series analysis.   Pharmacoepidemiol Drug Saf. 2017;26(9):1093-1099. doi:10.1002/pds.4240 PubMedGoogle ScholarCrossref
5.
Olfson  M, Druss  BG, Marcus  SC.  Trends in mental health care among children and adolescents.   N Engl J Med. 2015;372(21):2029-2038. doi:10.1056/NEJMsa1413512 PubMedGoogle ScholarCrossref
6.
Ghandour  RM, Sherman  LJ, Vladutiu  CJ,  et al. Prevalence and treatment of depression, anxiety, and conduct problems in US children.  J Pediatr. 2019;206:256-267.e3. doi:10.1016/j.jpeds.2018.09.021
7.
Mojtabai  R, Olfson  M, Han  B.  National trends in the prevalence and treatment of depression in adolescents and young adults.   Pediatrics. 2016;138(6):e20161878. doi:10.1542/peds.2016-1878 PubMedGoogle Scholar
8.
Bachmann  CJ, Aagaard  L, Burcu  M,  et al.  Trends and patterns of antidepressant use in children and adolescents from five western countries, 2005-2012.   Eur Neuropsychopharmacol. 2016;26(3):411-419. doi:10.1016/j.euroneuro.2016.02.001 PubMedGoogle ScholarCrossref
9.
Strawn  JR, Welge  JA, Wehry  AM, Keeshin  B, Rynn  MA.  Efficacy and tolerability of antidepressants in pediatric anxiety disorders: a systematic review and meta-analysis.   Depress Anxiety. 2015;32(3):149-157. doi:10.1002/da.22329 PubMedGoogle ScholarCrossref
10.
Walkup  JT.  Antidepressant efficacy for depression in children and adolescents: industry- and NIMH-funded studies.   Am J Psychiatry. 2017;174(5):430-437. doi:10.1176/appi.ajp.2017.16091059 PubMedGoogle ScholarCrossref
11.
Andersohn  F, Schade  R, Suissa  S, Garbe  E.  Long-term use of antidepressants for depressive disorders and the risk of diabetes mellitus.   Am J Psychiatry. 2009;166(5):591-598. doi:10.1176/appi.ajp.2008.08071065 PubMedGoogle ScholarCrossref
12.
Pan  A, Sun  Q, Okereke  OI,  et al.  Use of antidepressant medication and risk of type 2 diabetes: results from three cohorts of US adults.   Diabetologia. 2012;55(1):63-72. doi:10.1007/s00125-011-2268-4 PubMedGoogle ScholarCrossref
13.
Salvi  V, Grua  I, Cerveri  G, Mencacci  C, Barone-Adesi  F.  The risk of new-onset diabetes in antidepressant users: a systematic review and meta-analysis.   PLoS One. 2017;12(7):e0182088. doi:10.1371/journal.pone.0182088 PubMedGoogle Scholar
14.
Bhattacharjee  S, Bhattacharya  R, Kelley  GA, Sambamoorthi  U.  Antidepressant use and new-onset diabetes: a systematic review and meta-analysis.   Diabetes Metab Res Rev. 2013;29(4):273-284. doi:10.1002/dmrr.2393 PubMedGoogle ScholarCrossref
15.
Rubin  RR, Ma  Y, Marrero  DG,  et al; Diabetes Prevention Program Research Group.  Elevated depression symptoms, antidepressant medicine use, and risk of developing diabetes during the diabetes prevention program.   Diabetes Care. 2008;31(3):420-426. doi:10.2337/dc07-1827 PubMedGoogle ScholarCrossref
16.
Kivimäki  M, Hamer  M, Batty  GD,  et al.  Antidepressant medication use, weight gain, and risk of type 2 diabetes: a population-based study.   Diabetes Care. 2010;33(12):2611-2616. doi:10.2337/dc10-1187 PubMedGoogle ScholarCrossref
17.
Vimalananda  VG, Palmer  JR, Gerlovin  H,  et al.  Depressive symptoms, antidepressant use, and the incidence of diabetes in the Black Women’s Health Study.   Diabetes Care. 2014;37(8):2211-2217. doi:10.2337/dc13-2642 PubMedGoogle ScholarCrossref
18.
Brown  LC, Majumdar  SR, Johnson  JA.  Type of antidepressant therapy and risk of type 2 diabetes in people with depression.   Diabetes Res Clin Pract. 2008;79(1):61-67. doi:10.1016/j.diabres.2007.07.009 PubMedGoogle ScholarCrossref
19.
Kivimäki  M, Batty  GD, Jokela  M,  et al.  Antidepressant medication use and risk of hyperglycemia and diabetes mellitus: a noncausal association?   Biol Psychiatry. 2011;70(10):978-984. doi:10.1016/j.biopsych.2011.07.008 PubMedGoogle ScholarCrossref
20.
Mezuk  B, Johnson-Lawrence  V, Lee  H,  et al.  Is ignorance bliss? depression, antidepressants, and the diagnosis of prediabetes and type 2 diabetes.   Health Psychol. 2013;32(3):254-263. doi:10.1037/a0029014 PubMedGoogle ScholarCrossref
21.
Azevedo Da Silva  M, Dugravot  A, Balkau  B,  et al; D.E.S.I.R. Study Group.  Antidepressant medication use and trajectories of fasting plasma glucose, glycated haemoglobin, β-cell function and insulin sensitivity: a 9-year longitudinal study of the D.E.S.I.R. cohort.   Int J Epidemiol. 2015;44(6):1927-1940. doi:10.1093/ije/dyv153 PubMedGoogle ScholarCrossref
22.
Pyykkönen  AJ, Räikkönen  K, Tuomi  T, Eriksson  JG, Groop  L, Isomaa  B.  Depressive symptoms, antidepressant medication use, and insulin resistance: the PPP-Botnia Study.   Diabetes Care. 2011;34(12):2545-2547. doi:10.2337/dc11-0107 PubMedGoogle ScholarCrossref
23.
Arterburn  D, Sofer  T, Boudreau  DM,  et al.  Long-term weight change after initiating second-generation antidepressants.   J Clin Med. 2016;5(4):E48. doi:10.3390/jcm5040048 PubMedGoogle Scholar
24.
Blumenthal  SR, Castro  VM, Clements  CC,  et al.  An electronic health records study of long-term weight gain following antidepressant use.   JAMA Psychiatry. 2014;71(8):889-896. doi:10.1001/jamapsychiatry.2014.414 PubMedGoogle ScholarCrossref
25.
Gafoor  R, Booth  HP, Gulliford  MC.  Antidepressant utilisation and incidence of weight gain during 10 years’ follow-up: population based cohort study.   BMJ. 2018;361:k1951. doi:10.1136/bmj.k1951 PubMedGoogle ScholarCrossref
26.
Burcu  M, Zito  JM, Safer  DJ,  et al.  Association of antidepressant medications with incident type 2 diabetes among Medicaid-insured youths.   JAMA Pediatr. 2017;171(12):1200-1207. doi:10.1001/jamapediatrics.2017.2896 PubMedGoogle ScholarCrossref
27.
Burcu  M, Zito  JM, Safer  DJ,  et al.  Concomitant use of atypical antipsychotics with other psychotropic medication classes and the risk of type 2 diabetes mellitus.   J Am Acad Child Adolesc Psychiatry. 2017;56(8):642-651. doi:10.1016/j.jaac.2017.04.004 PubMedGoogle ScholarCrossref
28.
Mansur  RB, McIntyre  RS, Brietzke  E.  Mental disorders, antidepressants, and type 2 diabetes.   JAMA Pediatr. 2018;172(7):703-704. doi:10.1001/jamapediatrics.2018.1005 PubMedGoogle ScholarCrossref
29.
Suglia  SF, Demmer  RT, Wahi  R, Keyes  KM, Koenen  KC.  Depressive symptoms during adolescence and young adulthood and the development of type 2 diabetes mellitus.   Am J Epidemiol. 2016;183(4):269-276. doi:10.1093/aje/kwv149 PubMedGoogle ScholarCrossref
30.
Shomaker  LB, Tanofsky-Kraff  M, Stern  EA,  et al.  Longitudinal study of depressive symptoms and progression of insulin resistance in youth at risk for adult obesity.   Diabetes Care. 2011;34(11):2458-2463. doi:10.2337/dc11-1131 PubMedGoogle ScholarCrossref
31.
Pinhas-Hamiel  O, Zeitler  P.  Acute and chronic complications of type 2 diabetes mellitus in children and adolescents.   Lancet. 2007;369(9575):1823-1831. doi:10.1016/S0140-6736(07)60821-6 PubMedGoogle ScholarCrossref
32.
Ivanovska  V, Rademaker  CM, van Dijk  L, Mantel-Teeuwisse  AK.  Pediatric drug formulations: a review of challenges and progress.   Pediatrics. 2014;134(2):361-372. doi:10.1542/peds.2013-3225 PubMedGoogle ScholarCrossref
33.
McMahon  AW, Dal Pan  G.  Assessing drug safety in children—the role of real-world data.   N Engl J Med. 2018;378(23):2155-2157. doi:10.1056/NEJMp1802197 PubMedGoogle ScholarCrossref
34.
’t Jong  GW, Vulto  AG, de Hoog  M, Schimmel  KJ, Tibboel  D, van den Anker  JN.  Unapproved and off-label use of drugs in a children’s hospital.   N Engl J Med. 2000;343(15):1125. doi:10.1056/NEJM200010123431515 PubMedGoogle ScholarCrossref
35.
Frattarelli  DA, Galinkin  JL, Green  TP,  et al; American Academy of Pediatrics Committee on Drugs.  Off-label use of drugs in children.   Pediatrics. 2014;133(3):563-567. doi:10.1542/peds.2013-4060 PubMedGoogle ScholarCrossref
36.
Corny  J, Lebel  D, Bailey  B, Bussières  JF.  Unlicensed and off-label drug use in children before and after pediatric governmental initiatives.   J Pediatr Pharmacol Ther. 2015;20(4):316-328.PubMedGoogle Scholar
37.
Cheung  K, Teichert  M, Moll  HA, Stricker  BH, Visser  LE.  Filled prescriptions of age-related contraindicated drugs in children: a one-year nationwide cohort study in the Netherlands.   Int J Clin Pharm. 2018;40(5):1137-1143. doi:10.1007/s11096-018-0717-6 PubMedGoogle ScholarCrossref
38.
Hoon  D, Taylor  MT, Kapadia  P, Gerhard  T, Strom  BL, Horton  DB.  Trends in off-label drug use in ambulatory settings: 2006-2015.   Pediatrics. 2019;144(4):e20190896. doi:10.1542/peds.2019-0896 PubMedGoogle Scholar
39.
Lovegrove  MC, Weidle  NJ, Budnitz  DS.  Trends in emergency department visits for unsupervised pediatric medication exposures, 2004-2013.   Pediatrics. 2015;136(4):e821-e829. doi:10.1542/peds.2015-2092 PubMedGoogle ScholarCrossref
40.
Neubert  A, Dormann  H, Weiss  J,  et al.  The impact of unlicensed and off-label drug use on adverse drug reactions in paediatric patients.   Drug Saf. 2004;27(13):1059-1067. doi:10.2165/00002018-200427130-00006 PubMedGoogle ScholarCrossref
41.
Impicciatore  P, Choonara  I, Clarkson  A, Provasi  D, Pandolfini  C, Bonati  M.  Incidence of adverse drug reactions in paediatric in/out-patients: a systematic review and meta-analysis of prospective studies.   Br J Clin Pharmacol. 2001;52(1):77-83. doi:10.1046/j.0306-5251.2001.01407.x PubMedGoogle ScholarCrossref
42.
Lu  H, Rosenbaum  S.  Developmental pharmacokinetics in pediatric populations.   J Pediatr Pharmacol Ther. 2014;19(4):262-276.PubMedGoogle Scholar
43.
Sherman  RE, Anderson  SA, Dal Pan  GJ,  et al.  Real-world evidence—what is it and what can it tell us?   N Engl J Med. 2016;375(23):2293-2297. doi:10.1056/NEJMsb1609216 PubMedGoogle ScholarCrossref
44.
Osokogu  OU, Dukanovic  J, Ferrajolo  C,  et al.  Pharmacoepidemiological safety studies in children: a systematic review.   Pharmacoepidemiol Drug Saf. 2016;25(8):861-870. doi:10.1002/pds.4041 PubMedGoogle ScholarCrossref
45.
Schneeweiss  S, Avorn  J.  A review of uses of health care utilization databases for epidemiologic research on therapeutics.   J Clin Epidemiol. 2005;58(4):323-337. doi:10.1016/j.jclinepi.2004.10.012 PubMedGoogle ScholarCrossref
46.
Lund  JL, Richardson  DB, Stürmer  T.  The active comparator, new user study design in pharmacoepidemiology: historical foundations and contemporary application.   Curr Epidemiol Rep. 2015;2(4):221-228. doi:10.1007/s40471-015-0053-5 PubMedGoogle ScholarCrossref
47.
Patten  SB.  Bupropion confirmed not to cause weight gain in a real-world clinical population.   Evid Based Ment Health. 2015;18(2):57. doi:10.1136/eb-2014-101978 PubMedGoogle ScholarCrossref
48.
Cheung  AH, Zuckerbrot  RA, Jensen  PS, Laraque  D, Stein  REK; GLAD-PC STEERING GROUP.  Guidelines for Adolescent Depression in Primary Care (GLAD-PC): part II, treatment and ongoing management.   Pediatrics. 2018;141(3):e20174082. doi:10.1542/peds.2017-4082 PubMedGoogle Scholar
49.
Cipriani  A, Zhou  X, Del Giovane  C,  et al.  Comparative efficacy and tolerability of antidepressants for major depressive disorder in children and adolescents: a network meta-analysis.   Lancet. 2016;388(10047):881-890. doi:10.1016/S0140-6736(16)30385-3 PubMedGoogle ScholarCrossref
50.
Teltsch  DY, Fazeli Farsani  S, Swain  RS,  et al.  Development and validation of algorithms to identify newly diagnosed type 1 and type 2 diabetes in pediatric population using electronic medical records and claims data.   Pharmacoepidemiol Drug Saf. 2019;28(2):234-243. doi:10.1002/pds.4728 PubMedGoogle ScholarCrossref
51.
Desai  RJ, Rothman  KJ, Bateman  BT, Hernandez-Diaz  S, Huybrechts  KF.  A propensity-score-based fine stratification approach for confounding adjustment when exposure is infrequent.   Epidemiology. 2017;28(2):249-257. doi:10.1097/EDE.0000000000000595 PubMedGoogle ScholarCrossref
52.
Robins  J, Hernán  M, Fitzmaurice  G, Davidian  M, Verbeke  G, Molenberghs  G.  Longitudinal Data Analysis. Chapman and Hall/CRC Press; 2009:553-599.
53.
Hernán  MA, Lanoy  E, Costagliola  D, Robins  JM.  Comparison of dynamic treatment regimes via inverse probability weighting.   Basic Clin Pharmacol Toxicol. 2006;98(3):237-242. doi:10.1111/j.1742-7843.2006.pto_329.x PubMedGoogle ScholarCrossref
54.
Hernán  MA.  The hazards of hazard ratios.   Epidemiology. 2010;21(1):13-15. doi:10.1097/EDE.0b013e3181c1ea43 PubMedGoogle ScholarCrossref
55.
Murray  EJ, Hernán  MA.  Improved adherence adjustment in the Coronary Drug Project.   Trials. 2018;19(1):158. doi:10.1186/s13063-018-2519-5 PubMedGoogle ScholarCrossref
56.
Bobo  WV, Cooper  WO, Stein  CM,  et al.  Antipsychotics and the risk of type 2 diabetes mellitus in children and youth.   JAMA Psychiatry. 2013;70(10):1067-1075. doi:10.1001/jamapsychiatry.2013.2053 PubMedGoogle ScholarCrossref
57.
Galling  B, Roldán  A, Nielsen  RE,  et al.  Type 2 diabetes mellitus in youth exposed to antipsychotics: a systematic review and meta-analysis.   JAMA Psychiatry. 2016;73(3):247-259. doi:10.1001/jamapsychiatry.2015.2923 PubMedGoogle ScholarCrossref
58.
American Academy of Pediatrics Subcommittee on Attention-Deficit/Hyperactivity Disorder and Committee on Quality Improvement.  Clinical practice guideline: treatment of the school-aged child with attention-deficit/hyperactivity disorder.   Pediatrics. 2001;108(4):1033-1044. doi:10.1542/peds.108.4.1033 PubMedGoogle ScholarCrossref
59.
Cole  SR, Hernán  MA.  Constructing inverse probability weights for marginal structural models.   Am J Epidemiol. 2008;168(6):656-664. doi:10.1093/aje/kwn164 PubMedGoogle ScholarCrossref
60.
Schneeweiss  S, Rassen  JA, Glynn  RJ, Avorn  J, Mogun  H, Brookhart  MA.  High-dimensional propensity score adjustment in studies of treatment effects using health care claims data.   Epidemiology. 2009;20(4):512-522. doi:10.1097/EDE.0b013e3181a663cc PubMedGoogle ScholarCrossref
61.
Lipsitch  M, Tchetgen Tchetgen  E, Cohen  T.  Negative controls: a tool for detecting confounding and bias in observational studies.   Epidemiology. 2010;21(3):383-388. doi:10.1097/EDE.0b013e3181d61eeb PubMedGoogle ScholarCrossref
62.
Marmorstein  NR, Iacono  WG, Legrand  L.  Obesity and depression in adolescence and beyond: reciprocal risks.   Int J Obes (Lond). 2014;38(7):906-911. doi:10.1038/ijo.2014.19 PubMedGoogle ScholarCrossref
63.
Abbasi  A, Juszczyk  D, van Jaarsveld  CHM, Gulliford  MC.  Body mass index and incident type 1 and type 2 diabetes in children and young adults: a retrospective cohort study.   J Endocr Soc. 2017;1(5):524-537. doi:10.1210/js.2017-00044 PubMedGoogle ScholarCrossref
64.
Austin  PC.  Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.   Stat Med. 2009;28(25):3083-3107. doi:10.1002/sim.3697 PubMedGoogle ScholarCrossref
65.
Deuschle  M.  Effects of antidepressants on glucose metabolism and diabetes mellitus type 2 in adults.   Curr Opin Psychiatry. 2013;26(1):60-65. doi:10.1097/YCO.0b013e32835a4206 PubMedGoogle ScholarCrossref
66.
McIntyre  RS, Soczynska  JK, Konarski  JZ, Kennedy  SH.  The effect of antidepressants on glucose homeostasis and insulin sensitivity: synthesis and mechanisms.   Expert Opin Drug Saf. 2006;5(1):157-168. doi:10.1517/14740338.5.1.157 PubMedGoogle ScholarCrossref
67.
Dabelea  D, Mayer-Davis  EJ, Saydah  S,  et al; SEARCH for Diabetes in Youth Study.  Prevalence of type 1 and type 2 diabetes among children and adolescents from 2001 to 2009.   JAMA. 2014;311(17):1778-1786. doi:10.1001/jama.2014.3201 PubMedGoogle ScholarCrossref
68.
Gaskin  DJ, Thorpe  RJ  Jr, McGinty  EE,  et al.  Disparities in diabetes: the nexus of race, poverty, and place.   Am J Public Health. 2014;104(11):2147-2155. doi:10.2105/AJPH.2013.301420 PubMedGoogle ScholarCrossref
69.
Scott  RA, Langenberg  C, Sharp  SJ,  et al; InterAct Consortium.  The link between family history and risk of type 2 diabetes is not explained by anthropometric, lifestyle or genetic risk factors: the EPIC-InterAct study.   Diabetologia. 2013;56(1):60-69. doi:10.1007/s00125-012-2715-x PubMedGoogle ScholarCrossref
70.
Dabelea  D, Mayer-Davis  EJ, Lamichhane  AP,  et al.  Association of intrauterine exposure to maternal diabetes and obesity with type 2 diabetes in youth: the SEARCH Case-Control Study.   Diabetes Care. 2008;31(7):1422-1426. doi:10.2337/dc07-2417 PubMedGoogle ScholarCrossref
71.
Ferguson  JM.  SSRI antidepressant medications: adverse effects and tolerability.   Prim Care Companion J Clin Psychiatry. 2001;3(1):22-27. doi:10.4088/PCC.v03n0105 PubMedGoogle ScholarCrossref
72.
Shatin  D, Levin  R, Ireys  HT, Haller  V.  Health care utilization by children with chronic illnesses: a comparison of Medicaid and employer-insured managed care.   Pediatrics. 1998;102(4):E44. doi:10.1542/peds.102.4.e44 PubMedGoogle Scholar
73.
Kelsey  JL, Whittemore  AS, Evans  AS, Thompson  WD.  Methods in Observational Epidemiology. Monographs in Epidemiology and Biostatistics; 1996.
74.
Huitfeldt  A, Hernan  MA, Kalager  M, Robins  JM.  Comparative effectiveness research using observational data: active comparators to emulate target trials with inactive comparators.   EGEMS (Wash DC). 2016;4(1):1234. doi:10.13063/2327-9214.1234 PubMedGoogle Scholar
75.
Boudreau  DM, Arterburn  D, Bogart  A,  et al.  Influence of body mass index on the choice of therapy for depression and follow-up care.   Obesity (Silver Spring). 2013;21(3):E303-E313. doi:10.1002/oby.20048 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    EXPAND ALL
    Comment on: Association of Selective Serotonin Reuptake Inhibitors With the Risk of Type 2 Diabetes in Children and Adolescents
    Liang Xu, Master of Medicine | Department of Geriatric Psychiatry, Huzhou Third Municipal Hospital, the Affiliated Hospital of Huzhou University, Huzhou, Zhejiang, China
    The researchers tried to examine the association between SSRIs and Type 2 Diabetes (T2D), with careful control for potential biases. However, I would like to raise some issues that should be clarified.
    First, a small increased risk of T2D associated with SSRI treatment was observed in children and adolescents among publicly insured patients, whereas the increased risk was not observed in privately insured patients. The researchers did not analyze the reasons for different underlying risks for T2D in-depth, only described the result as the main conclusion. Compared with privately insured patients, the publicly insured patients were of lower socioeconomic status,
    had more comorbidities, and were in a higher prevalence of risk factors for T2D.1 So the Medicaid Analytic eXtract Population (publicly insured) had a higher incidence rate of T2D (eTable10 in the Supplement). Theoretically, in the Medicaid Analytic eXtract Population, a cohort with more confounding factors and greater confounding intensity(low income, comorbidities, eating habits, daily exercise, health awareness, etc.), the SSRI treatment should have a weaker effect on the T2D compared in the IBM MarketScan Population.
    However, the result was just the opposite. Although more than 100 covariates were selected as confounders or proxies of confounders, there were still some covariables(family history of T2D, eating habits, daily exercise, health awareness, etc.) that could not be directly controlled, resulting in different risks for T2D. Further comprehensive analysis of the differences between the two populations might help understand why the risk for T2D was different between the two populations and provide more information on the association of SSRI with the risk of T2D in Children and Adolescents.
    Second, the researchers examined the associations between SSRIs and incidence of hyperlipidemia and asthma, expecting a null finding to provide indirect evidence of no substantial residual confounding or detection bias. But in the sensitivity analyses in ITT population(eTable 17), the adjusted hazard ratio(95% CI) of asthma and hyperlipidemia is 0.97 (0.95-0.99), 0.96 (0.92-1.00), respectively in MAX; 0.97 (0.94-1.01), 0.83 (0.78-0.88), respectively in MarketScan. It showed that residual confounders were present, which may influence these results.



    References:
    1.Shatin D, Levin R, Ireys HT, Haller V. Health care utilization by children with chronic illnesses: a comparison of Medicaid and employer-insured managed care. Pediatrics 1998;102(4):e44-e44
    CONFLICT OF INTEREST: None Reported
    READ MORE
    Original Investigation
    September 2, 2020

    Association of Selective Serotonin Reuptake Inhibitors With the Risk of Type 2 Diabetes in Children and Adolescents

    Author Affiliations
    • 1Department of Epidemiology, Harvard T. H. Chan School of Public Health, Boston, Massachusetts
    • 2Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
    • 3Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, Massachusetts
    • 4Department of Pediatrics, Harvard Medical School, Boston, Massachusetts
    • 5Department of Psychiatry, Columbia University Vagelos College of Physicians and Surgeons, New York, New York
    • 6Department of Anesthesiology, Perioperative, and Pain Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
    JAMA Psychiatry. 2021;78(1):91-100. doi:10.1001/jamapsychiatry.2020.2762
    Key Points

    Question  Is the treatment of children and adolescents with selective serotonin reuptake inhibitors (SSRIs) associated with an increased risk of type 2 diabetes?

    Findings  In a US nationwide pediatric cohort study of more than 1 million publicly and privately insured patients, initiation of SSRI treatment was associated with a small increased risk of type 2 diabetes among publicly insured patients, corresponding to a number needed to harm of 1515 treated for at least 2 years for 1 patient to develop type 2 diabetes.

    Meaning  Children and adolescents who initiate SSRI treatment may be at a small increased risk of developing type 2 diabetes, but the magnitude was more modest than previously reported, and the absolute risk was small.

    Abstract

    Importance  Concerns exist that use of selective serotonin reuptake inhibitors (SSRIs) increases the risk of developing type 2 diabetes (T2D) in adults, but evidence in children and adolescents is limited. In the absence of a randomized clinical trial, evidence must be generated using real-world data.

    Objective  To evaluate the safety of SSRI use in children and adolescents with respect to the associated risk of T2D.

    Design, Setting, and Participants  This cohort study of patients aged 10 to 19 years with a diagnosis for an SSRI treatment indication was conducted within the nationwide Medicaid Analytic eXtract (MAX; January 1, 2000, to December 31, 2014) and the IBM MarketScan (January 1, 2003, to September 30, 2015) databases. Data were analyzed from November 1, 2018, to December 6, 2019.

    Exposures  New users of an SSRI medication and comparator groups with no known metabolic adverse effects (no antidepressant exposure, bupropion hydrochloride exposure, or psychotherapy exposure). Within-class individual SSRI medications were compared with fluoxetine hydrochloride.

    Main Outcomes and Measures  Incident T2D during follow-up. Intention-to-treat effects were estimated using Cox proportional hazards regression models, adjusting for confounding through propensity score stratification. As-treated effects to account for continuous treatment were estimated using inverse probability weighting and marginal structural models.

    Results  A total of 1 582 914 patients were included in the analysis (58.3% female; mean [SD] age, 15.1 [2.3] years). The SSRI-treated group included 316 178 patients in the MAX database (publicly insured; mean [SD] age, 14.7 [2.1] years; 62.2% female) and 211 460 in the MarketScan database (privately insured; mean [SD] age, 15.8 [2.3] years; 63.9% female) with at least 2 SSRI prescriptions filled, followed up for a mean (SD) of 2.3 (2.0) and 2.2 (1.9) years, respectively. In publicly insured patients, initiation of SSRI treatment was associated with a 13% increased hazard of T2DM (intention-to-treat adjusted hazard ratio [aHR], 1.13; 95% CI, 1.04-1.22) compared with untreated patients. The association strengthened for continuous SSRI treatment (as-treated aHR, 1.33; 95% CI, 1.21-1.47), corresponding to 6.6 (95% CI, 4.2-10.4) additional cases of T2D per 10 000 patients treated for at least 2 years. Adjusted HRs were lower in privately insured patients (intention-to-treat aHR, 1.01 [95% CI, 0.84-1.23]; as-treated aHR, 1.10 [95% CI, 0.88-1.36]). Findings were similar when comparing SSRI treatment with psychotherapy (publicly insured as-treated aHR, 1.44 [95% CI, 1.25-1.65]; privately insured as-treated aHR, 1.21 [95% CI, 0.93-1.57]), whereas no increased risk was observed compared with bupropion treatment publicly insured as-treated aHR, 1.01 [95% CI, 0.79-1.29]; privately insured as-treated aHR, 0.87 [95% CI, 0.44-1.70]). For the within-class analysis, no medication had an increased hazard of T2D compared with fluoxetine.

    Conclusions and Relevance  These findings suggest that children and adolescents initiating SSRI treatment may be at a small increased risk of developing T2D, particularly publicly insured patients. The magnitude of association was more modest than previously reported, and the absolute risk was small. The potential small risk should be viewed in relation to the efficacy of SSRIs for its major indications in young patients.

    Introduction

    Children and adolescents taking selective serotonin reuptake inhibitors (SSRIs) should be closely monitored, given the potential adverse effects.1-3 Despite these concerns, SSRI use in young patients remains widespread, given the increasing prevalence of depression in this population4-8 and the efficacy of SSRIs for pediatric depression and anxiety disorders.9,10 In 2012, an estimated 3.5% of children (aged 10-14 years) and 6.2% of adolescents (aged 15-19 years) in the United States were taking an antidepressant, with 72% of these patients treated with an SSRI.8

    An association between SSRIs and type 2 diabetes (T2D) has been reported in several studies conducted in adults.11-18 Although residual biases could not be excluded as alternative explanations,19-22 weight gain is a potential biological mechanism. Selective serotonin reuptake inhibitors are generally associated with modest weight gain,23-25 but this may be sufficient in transitioning some patients from normal weight to overweight/obesity and elevate the risk of T2D.25 Evidence regarding this potential safety concern in children and adolescents is limited.26,27 A recent observational study26 found that treatment with SSRIs or serotonin and norepinephrine reuptake inhibitors may be associated with a 90% increased risk of T2D in youths (aged 5-20 years). However, this study was not sufficiently large to examine the SSRI class alone or specific SSRIs, which is needed to inform clinical decision-making, because weight gain and metabolic adverse effects differ by class and for individual drugs within the same class.25 Furthermore, it is possible that behaviors associated with depression severity may have contributed to the observed risk.28-30 More evidence is needed, given the widespread prescription of SSRIs and potential health threats posed by T2D onset during childhood and adolescence.31

    Children and adolescents are often excluded from randomized clinical trials (RCTs), resulting in a lack of evidence on medication safety and widespread off-label prescribing.32-41 Rapid changes in growth during childhood and adolescence can alter drugs’ pharmacokinetics and pharmacodynamics, so high-quality age-specific drug safety data are needed to inform prescribing decisions.42 Because it is unlikely that a sufficiently large RCT will be conducted to study SSRIs and T2D in young patients, evidence must be generated using real-world data.33,43,44 Using data from 2 nationwide health care databases, we conducted a cohort study of children and adolescents to examine the association between SSRIs and T2D, with careful control for potential biases.

    Methods
    Study Population

    This study leveraged nationwide US claims data from the Medicaid Analytic eXtract (MAX; January 1, 2000, to December 31, 2014) database, consisting of patients enrolled in Medicaid and the Children’s Health Insurance Program, and IBM MarketScan (MarketScan; January 1, 2003, to September 30, 2015) database, consisting of privately insured patients. Within these databases, we identified patients aged 10 to 19 years with a diagnosis for an SSRI indication (ie, depression, generalized or social anxiety disorder, obsessive-compulsive disorder, posttraumatic stress disorder, panic disorder, or bulimia nervosa) (eTable 1 in Supplement). The research was approved by the institutional review board of Brigham and Women’s Hospital, which granted a waiver of informed consent for the use of deidentified data. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

    Exposure

    The exposed group consisted of new users of an SSRI (eTable 2 in the Supplement). The index date was defined as the date of the first observed SSRI dispensing. Patients without continuous enrollment and those with evidence of antidepressant use, diabetes-related conditions, pregnancy, hospice care, or serious medical conditions during the 365 days before the index date were excluded (eTable 1 in the Supplement).

    We compared patients who initiated SSRI treatment with patients who had a diagnosis for an SSRI indication but no antidepressant exposure (untreated group). At the time of each SSRI user’s index date, we randomly selected 2 untreated patients who had at least 1 medical encounter within 14 days before or after the index date and were the same age. The untreated patient’s index date was the medical encounter date. In secondary analyses, 2 active comparators with no known metabolic adverse effects were used to further reduce the risk of confounding by the indication or its severity.45,46 Specifically, we identified new users of bupropion hydrochloride, an antidepressant known to not result in weight gain,23,24,47 and new users of psychotherapy. The index dates were defined as the date of the first bupropion dispensing and the first health care claim for psychotherapy, respectively. Patients were required to satisfy the previously outlined inclusion and exclusion criteria.

    We also compared the individual SSRI medications (within-class comparison).45 Patients receiving fluoxetine hydrochloride served as the reference group.48,49 Classification of patients into treatment groups (citalopram hydrobromide, escitalopram oxalate, fluoxetine, fluvoxamine maleate, paroxetine hydrochloride or paroxetine mesylate, and sertraline hydrochloride) was based on the index dispensing.

    Outcome

    Patients were followed up until the first diagnosis of T2D. Cases were identified using a validated algorithm developed to identify T2D in children based on administrative data (sensitivity, 90.3%; specificity, 95.0%).50 This definition was based on the presence of inpatient or outpatient diagnoses for T2D and use of antidiabetic medications. Patients were censored at the time of other diabetes-related medical encounters (type 1, gestational, and secondary diabetes), hospice care, pregnancy, end of insurance coverage, or end of available data.

    Covariates

    Baseline characteristics were measured during the 365 days before the index date. More than 100 covariates were selected a priori as plausible confounders or proxies of confounders. These include demographic characteristics, psychiatric diagnoses, metabolic conditions, concomitant medications, and use of health care services (eTable 3 in the Supplement).

    Intention-to-Treat Analysis

    To minimize exposure misclassification, the intention-to-treat (ITT) analysis was restricted to patients with at least 1 additional SSRI prescription during the 180 days after the index dispensing (exposure assessment period). Untreated patients were required to have no antidepressant prescriptions and at least 1 additional medical encounter during the exposure assessment period. Active comparators were similarly constructed (eFigure 1 in the Supplement). Patients with the outcome or a censorship event during the exposure assessment period were excluded. Follow-up began the 181st day after the index date. Consistent with an ITT approach, we disregarded treatment changes during follow-up.

    To adjust for baseline confounding, we used propensity score stratification. After trimming observations from the nonoverlapping regions of the propensity score distribution, we constructed 10 propensity score strata based on the SSRI group’s propensity-score distribution. Within the strata of propensity score, patients in the comparator group were weighted according to the distribution of the SSRI exposed.51 Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs.

    As-Treated Analysis

    An as-treated analysis was conducted to estimate the association of continuous SSRI treatment (vs untreated, bupropion treatment, and psychotherapy). For each 3-month window of follow-up (after the index date), we assessed whether patients remained adherent to their index exposure group. Specifically, SSRI and bupropion users were required to have a supply of at least 30 days for the respective treatment, and untreated patients and those receiving psychotherapy were required to remain free of antidepressant exposure. Patients were censored at the end of the 3-month window during which they no longer satisfied the definition for their baseline treatment group (ie, discontinued), initiated use of a different antidepressant, or had a previously described study censorship event. This analysis had a maximum 5-year observation period, because few patients remained in the risk set beyond that point.

    Inverse probability of treatment weights, estimated using baseline covariates, were used to account for baseline confounding. Inverse probability of censoring weights, estimated as a function of baseline and time-varying covariates (updated every 3 months), were used to adjust for confounding and selection bias due to censoring.52,53 Weights were truncated at the 1st and 99th percentiles to prevent outliers from affecting the analysis. Pooled logistic marginal structural models were used to estimate HRs, and robust variance estimators were used to compute 95% CIs.54,55 In sensitivity analyses, we considered different approaches to modeling and truncating weights.

    Subgroup and Sensitivity Analyses

    Several subgroup and sensitivity analyses were conducted to evaluate the robustness of our primary results (between-class) in the ITT population. To explore potential effect modification, subgroup analyses were conducted separately for age (10-14 and 15-19 years), sex (female and male), and race/ethnicity (non-Hispanic White, Black, Hispanic/Latino, and other [MAX database only]). Additional subgroups were restricted to depression and anxiety diagnoses (the most common SSRI indications), excluded antipsychotic users (risk factor for T2D),56,57 and were restricted to stimulant users (could limit weight gain).58

    In sensitivity analyses, the high-dimensional propensity score was used to account for proxies of unmeasured confounding.59,60 In negative control analyses, we examined the associations between SSRIs and incidence of hyperlipidemia and asthma. Control outcomes were selected a priori as outcomes that should have no association with SSRIs but may have a similar structure of confounding or a similar potential for detection bias as T2D. A null finding would provide indirect evidence of no substantial residual confounding or detection bias.61

    We also conducted sensitivity analyses to explore potential residual confounding by obesity, which is poorly captured in administrative data but could be an important confounder, because adolescents with more severe depression are more likely to receive SSRIs, to have overweight/obesity, and to develop T2D.29,30,62,63 For a subset of the MAX population, body mass index data were obtained by linking their claims data to the Partners Healthcare Research Patient Data Registry, an electronic health record database consisting of patients who receive care in the Partners hospital systems in Massachusetts. Details are given in the eMethods in the Supplement.

    Data were analyzed from November 1, 2018, to December 6, 2019. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc).

    Results
    Patient Characteristics

    A total of 1 582 914 patients were included in the primary ITT analysis (58.3% female; 41.7% male; mean [SD] age, 15.1 [2.3] years). The ITT cohort included 316 178 patients who initiated SSRI treatment and 632 356 age-matched untreated patients in the MAX database (publicly insured) and 211 460 patients who initiated SSRI treatment and 422 920 age-matched untreated patients in the MarketScan database (privately insured). The as-treated cohort was larger (≥2 prescriptions not required), with 439 205 SSRI-treated and 878 410 age-matched untreated publicly insured patients and 269 762 SSRI-treated and 539 524 age-matched untreated privately insured patients.

    The SSRI-treated patients were more likely to be female (62.9%) and white (61.8%) (publicly insured only). They also had more medical diagnoses and medication use for both psychiatric-related and unrelated conditions (eg, 10.5% SSRI-treated vs 7.3% untreated had a diagnosis of bipolar disorder; 11.0% SSRI-treated vs 7.3% untreated filled a prescription for antipsychotics). Markers for obesity and metabolic conditions were similar between treated and untreated patients. After propensity-score stratification, all baseline characteristics were balanced across treatment groups, with absolute standardized differences of less than 0.10 (Table 1 and eTables 4 and 5 in the Supplement).64 Use of active comparators (bupropion and psychotherapy) also achieved balance in characteristics (eTables 6-9 in the Supplement).

    In the ITT cohort, patients who initiated SSRI treatment were followed up for a mean (SD) of 2.3 (2.0) years in the publicly insured and 2.2 (1.9) years in the privately insured cohorts. Mean (SD) follow-up was shorter in the as-treated population (SSRI treated: 1.0 [1.0] years in publicly insured cohort and 1.1 [1.1] years in privately insured cohort) (Table 2 and eTable 10 in the Supplement).

    Between-Class and Within-Class Comparisons

    In publicly insured patients, the unadjusted incidence rate of T2D was 2.32 cases per 1000 person-years among patients who initiated SSRI treatment and 1.65 cases per 1000 person-years among untreated patients in the ITT population (Table 2). Unadjusted HRs in ITT analyses suggested an association between SSRI initiation and T2D compared with noninitiators (crude HR, 1.39; 95% CI, 1.31-1.49). After fully adjusting for baseline covariates, the association moved toward the null (adjusted HR [aHR], 1.13; 95% CI, 1.04-1.22) (Figure 1 and eTable 11 in the Supplement). The magnitude of association increased for continuous SSRI treatment in as-treated analyses (aHR, 1.33; 95% CI, 1.21-1.47), which corresponded to an additional 6.6 (95% CI, 4.2-10.4) T2D cases per 10 000 patients continuously treated for at least 2 years (number needed to harm, 1515 patients treated for ≥2 years for 1 patient to develop T2D) and an additional 28.4 (95%, 16.0-46.3) T2D cases per 10 000 continuously treated for at least 5 years (number needed to harm, 352) (eTable 12 in the Supplement). The association did not manifest until more than 1 year of follow-up (Figure 2).

    In privately insured patients, absolute rates in the ITT population were lower: 0.63 cases per 1000 person-years among patients who initiated SSRI treatment and 0.54 cases per 1000 person-years among untreated patients (Table 2). After adjustment, there was no increased hazard of T2D in ITT (aHR, 1.01; 95% CI, 0.84-1.23) and no meaningful increased hazard of T2D in as-treated analyses (aHR, 1.10; 95% CI, 0.88-1.36) (Figure 1 and eTable 11 in the Supplement). This point estimate corresponds to an additional 3.7 (95% CI, −3.2 to 15.5) cases per 10 000 continuously treated for at least 5 years (eTable 12 in the Supplement).

    Using active comparators, no increased hazard was observed comparing SSRI treatment with bupropion treatment (publicly insured as-treated aHR, 1.01 [95% CI, 0.79-1.29]; privately insured as-treated aHR, 0.87 [95% CI, 0.44-1.70]) (eTable 13 in the Supplement), whereas an increased hazard was observed compared with psychotherapy (publicly insured as-treated aHR, 1.44 [95% CI, 1.25-1.65]; privately insured as-treated aHR, 1.21 [95% CI, 0.93-1.57]) (eTable 14 in the Supplement).

    Within-class HRs did not vary substantially by individual SSRI medication (Figure 3 and eTable 15 in the Supplement).

    Subgroup and Sensitivity Analyses

    Overall, findings did not vary substantially across subgroups (Figure 1 and eTable 16 in the Supplement). Results remained consistent after using high-dimensional propensity-score adjustment. No increased risks were observed in asthma and hyperlipidemia control outcomes, providing indirect evidence of no residual biases (eTable 17 in the Supplement). In as-treated analyses, different approaches to modeling and truncating weights did not alter our findings (eTable 18 in the Supplement). In bias analyses to explore the effect of residual confounding due to poorly measured obesity, we found that better measurement of obesity would not have substantially altered our conclusions (eMethods, eTables 19-22, and eFigure 2 in the Supplement).

    Discussion

    Using nationwide data from 2 US health care databases, we observed a small increased risk of T2D associated with SSRI treatment in children and adolescents, particularly among publicly insured patients. This association was not apparent during the first year of treatment and strengthened slightly with longer follow-up. Within the SSRI class, no individual medication increased the risk of T2D compared with fluoxetine. Findings were generally consistent across subgroup and sensitivity analyses.

    A previous study26 found that use of SSRIs or serotonin and norepinephrine reuptake inhibitors in Medicaid-insured youths was associated with a substantial increased risk of T2D (risk ratio, 1.88; 95% CI, 1.34-2.64), with the risk intensifying with longer duration (>150 days). Our study found an attenuated association among Medicaid-insured SSRI users after evaluating the SSRI class alone and using different techniques to control for confounding by indication (ie, restriction to patients with an SSRI indication, time-varying confounding adjustment when evaluating continuous treatment, active comparators). Furthermore, we expand the evidence base by examining privately insured SSRI users, who had no meaningful increased risk of T2D.

    Although small RCTs of adults65,66 found that SSRI treatment does not impair glucose metabolism in the short term, SSRI-induced weight gain may be associated with transitioning patients with normal weight to overweight/obesity25 and thereby elevating their risk for T2D. Unfortunately, the association between SSRIs and weight gain could not be directly assessed in our data. The long-term metabolic effects of SSRI treatment remain poorly characterized, with mixed evidence regarding the association between antidepressants and T2D, insulin resistance, and intermediate markers of T2D.11-22 In children and adolescents, evidence regarding SSRIs and glucose metabolism is limited. We found a small increased risk of T2D among publicly insured patients who initiated SSRI treatment. The association strengthened with longer treatment periods, but the magnitude was more modest than previously reported26 and substantially smaller than the risk associated with other known risk factors for T2D, such as obesity, race, and poverty (2- to 7-fold increased risk).56,63,67-70 When making treatment decisions in young patients, this potential small risk of T2D, along with other potential adverse effects of SSRIs such as nausea and sleep disturbances,2,3,71 should be weighed against the benefits of treating pediatric depression and anxiety disorders.9,10 Because the patients in our cohorts were free of T2D at baseline, our study does not inform the treatment of patients with established T2D.

    Heterogeneity in point estimates was observed across data sources. This is not unexpected, because the data sources reflect different populations with different underlying risks for T2D, comparable to heterogeneity observed in stratified analyses in RCTs. The increased risk of T2D was observed in publicly insured patients who are of lower socioeconomic status and represent a population with greater overall medical burden, more comorbidities, and a higher prevalence of risk factors for T2D,72 whereas no meaningful increased risk was observed in privately insured patients. These findings highlight the value of examining results across different populations with different underlying risks. Patients with high-risk characteristics should be monitored more closely for glucose intolerance.

    This study highlights the opportunities of using RWD to generate evidence on drug safety in children and adolescents. First, health care databases enabled large cohorts with sufficient power to examine the SSRI class and individual SSRI medications and to perform clinically relevant subgroup analyses. Second, the longitudinal nature of these databases provided a chance to evaluate the long-term effects of treatment and to capture the development of T2D.

    Limitations

    There are challenges to using RWD to evaluate drug safety, but approaches can be leveraged to improve its validity. First, as in all nonrandomized studies, we cannot rule out the possibility of residual confounding by unmeasured characteristics, such as control of baseline blood glucose levels. Several attempts were made to minimize this possibility, including use of active comparators, adjustment for at least 100 baseline covariates (including proxies for high blood glucose levels), further adjustment for time-varying confounding in as-treated analyses, and use of control outcomes. In addition, we conducted sensitivity analyses in a claims- and electronic health records–linked subset and found that body mass index was well-balanced between treatment groups, reducing its potential for confounding. However, in this linked subset, the sample size was small for bupropion. The potential role of confounding by obesity on observed estimates was illustrated in eFigure 2 in the Supplement.

    Second, RWD studies need evidence of exposure and a highly specific outcome definition.73 To increase the likelihood that patients took their medication, we required at least 1 additional prescription within the first 6 months after treatment initiation in ITT analyses. We also conducted as-treated analyses, censoring patients when they no longer received refills. Outcome misclassification was minimized by using a validated definition with high sensitivity and specificity.50

    Finally, point estimates varied across comparator groups. If we conducted an RCT, SSRI users would be compared with a placebo group. However, placebo groups are difficult to mimic in RWD,74 so we explored several options to broaden our understanding. A positive association was observed compared with the untreated and psychotherapy groups. However, untreated patients and those receiving psychotherapy likely differ from those who receive an antidepressant with respect to disease severity and other clinical factors. Compared with bupropion, for which confounding by indication may pose a lower threat to validity, we observed no increased risk. One additional explanation could be that patients with overweight/obesity were more likely to initiate bupropion treatment compared with other antidepressants.75 Given the lack of a known biological mechanism, further investigation is needed to explore the association between bupropion and T2D. Most importantly, across all comparisons, the magnitude of association was lower than initially reported, suggesting that the risk of T2D is less of a safety concern than previously proposed.

    Conclusions

    This cohort study found that treatment with SSRIs during childhood and adolescence may be associated with a small increased risk of T2D, particularly among publicly insured patients. The magnitude of association was more modest than previously reported, and the absolute risk was small, providing evidence that this safety concern is not as substantial as initially reported. This potential risk, which is much lower in magnitude than the other known risk factors for T2D, should be weighed against the known benefits and risks of SSRI treatment to help inform treatment decision making in the pediatric population.

    Back to top
    Article Information

    Accepted for Publication: July 5, 2020.

    Corresponding Author: Jenny W. Sun, PhD, Department of Epidemiology, Harvard T. H. Chan School of Public Health, 677 Huntington Ave, Boston, MA 02115 (jennysun@mail.harvard.edu).

    Published Online: September 2, 2020. doi:10.1001/jamapsychiatry.2020.2762

    Author Contributions: Dr Sun had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Sun, Olfson, Bateman, Huybrechts.

    Acquisition, analysis, or interpretation of data: Sun, Hernández-Díaz, Haneuse, Bourgeois, Vine, Bateman, Huybrechts.

    Drafting of the manuscript: Sun.

    Critical revision of the manuscript for important intellectual content: Hernández-Díaz, Haneuse, Bourgeois, Vine, Olfson, Bateman, Huybrechts.

    Statistical analysis: Sun, Haneuse, Vine.

    Administrative, technical, or material support: Huybrechts.

    Supervision: Hernández-Díaz, Bourgeois, Huybrechts.

    Conflict of Interest Disclosures: Dr Hernández-Díaz reported participating as an investigator in projects funded by GlaxoSmithKline and Eli Lilly and Company and consulting for Boehringer Ingelheim, Roche, and UCB as a methods advisor for pregnancy studies. Ms Vine reported previous employment by an independent consulting firm that worked with pharmaceutical companies for work unrelated to the topic of this study. Dr Bateman reported participating as an investigator in grants to Brigham and Women’s Hospital from Baxalta, GlaxoSmithKline, Pacira BioSciences, Inc, and Eli Lilly and Company; serving on an expert panel on a postpartum hemorrhage quality improvement project for mothers sponsored by a grant from Merck and Co; and consulting for the Alosa Foundation and Aetion Inc. Dr Huybrechts reported participating as an investigator in grants to the Brigham and Women’s Hospital from Eli Lilly and Company and GlaxoSmithKline. No other disclosures were reported.

    Funding/Support: This study was supported by a training grant from the Program in Pharmacoepidemiology at Harvard T. H. Chan School of Public Health (Dr Sun).

    Role of the Funder/Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    References
    1.
    Dwyer  JB, Bloch  MH.  Antidepressants for pediatric patients.   Curr Psychiatr. 2019;18(9):26-42F.PubMedGoogle Scholar
    2.
    Mills  JA, Strawn  JR.  Antidepressant tolerability in pediatric anxiety and obsessive-compulsive disorders: a Bayesian hierarchical modeling meta-analysis.   J Am Acad Child Adolesc Psychiatry. 2019;S0890-8567(19)32113-6. doi:10.1016/j.jaac.2019.10.013 PubMedGoogle Scholar
    3.
    Dobson  ET, Strawn  JR.  Pharmacotherapy for pediatric generalized anxiety disorder: a systematic evaluation of efficacy, safety and tolerability.   Paediatr Drugs. 2016;18(1):45-53. doi:10.1007/s40272-015-0153-1 PubMedGoogle ScholarCrossref
    4.
    Morkem  R, Williamson  T, Patten  S,  et al.  Trends in antidepressant prescribing to children and adolescents in Canadian primary care: a time-series analysis.   Pharmacoepidemiol Drug Saf. 2017;26(9):1093-1099. doi:10.1002/pds.4240 PubMedGoogle ScholarCrossref
    5.
    Olfson  M, Druss  BG, Marcus  SC.  Trends in mental health care among children and adolescents.   N Engl J Med. 2015;372(21):2029-2038. doi:10.1056/NEJMsa1413512 PubMedGoogle ScholarCrossref
    6.
    Ghandour  RM, Sherman  LJ, Vladutiu  CJ,  et al. Prevalence and treatment of depression, anxiety, and conduct problems in US children.  J Pediatr. 2019;206:256-267.e3. doi:10.1016/j.jpeds.2018.09.021
    7.
    Mojtabai  R, Olfson  M, Han  B.  National trends in the prevalence and treatment of depression in adolescents and young adults.   Pediatrics. 2016;138(6):e20161878. doi:10.1542/peds.2016-1878 PubMedGoogle Scholar
    8.
    Bachmann  CJ, Aagaard  L, Burcu  M,  et al.  Trends and patterns of antidepressant use in children and adolescents from five western countries, 2005-2012.   Eur Neuropsychopharmacol. 2016;26(3):411-419. doi:10.1016/j.euroneuro.2016.02.001 PubMedGoogle ScholarCrossref
    9.
    Strawn  JR, Welge  JA, Wehry  AM, Keeshin  B, Rynn  MA.  Efficacy and tolerability of antidepressants in pediatric anxiety disorders: a systematic review and meta-analysis.   Depress Anxiety. 2015;32(3):149-157. doi:10.1002/da.22329 PubMedGoogle ScholarCrossref
    10.
    Walkup  JT.  Antidepressant efficacy for depression in children and adolescents: industry- and NIMH-funded studies.   Am J Psychiatry. 2017;174(5):430-437. doi:10.1176/appi.ajp.2017.16091059 PubMedGoogle ScholarCrossref
    11.
    Andersohn  F, Schade  R, Suissa  S, Garbe  E.  Long-term use of antidepressants for depressive disorders and the risk of diabetes mellitus.   Am J Psychiatry. 2009;166(5):591-598. doi:10.1176/appi.ajp.2008.08071065 PubMedGoogle ScholarCrossref
    12.
    Pan  A, Sun  Q, Okereke  OI,  et al.  Use of antidepressant medication and risk of type 2 diabetes: results from three cohorts of US adults.   Diabetologia. 2012;55(1):63-72. doi:10.1007/s00125-011-2268-4 PubMedGoogle ScholarCrossref
    13.
    Salvi  V, Grua  I, Cerveri  G, Mencacci  C, Barone-Adesi  F.  The risk of new-onset diabetes in antidepressant users: a systematic review and meta-analysis.   PLoS One. 2017;12(7):e0182088. doi:10.1371/journal.pone.0182088 PubMedGoogle Scholar
    14.
    Bhattacharjee  S, Bhattacharya  R, Kelley  GA, Sambamoorthi  U.  Antidepressant use and new-onset diabetes: a systematic review and meta-analysis.   Diabetes Metab Res Rev. 2013;29(4):273-284. doi:10.1002/dmrr.2393 PubMedGoogle ScholarCrossref
    15.
    Rubin  RR, Ma  Y, Marrero  DG,  et al; Diabetes Prevention Program Research Group.  Elevated depression symptoms, antidepressant medicine use, and risk of developing diabetes during the diabetes prevention program.   Diabetes Care. 2008;31(3):420-426. doi:10.2337/dc07-1827 PubMedGoogle ScholarCrossref
    16.
    Kivimäki  M, Hamer  M, Batty  GD,  et al.  Antidepressant medication use, weight gain, and risk of type 2 diabetes: a population-based study.   Diabetes Care. 2010;33(12):2611-2616. doi:10.2337/dc10-1187 PubMedGoogle ScholarCrossref
    17.
    Vimalananda  VG, Palmer  JR, Gerlovin  H,  et al.  Depressive symptoms, antidepressant use, and the incidence of diabetes in the Black Women’s Health Study.   Diabetes Care. 2014;37(8):2211-2217. doi:10.2337/dc13-2642 PubMedGoogle ScholarCrossref
    18.
    Brown  LC, Majumdar  SR, Johnson  JA.  Type of antidepressant therapy and risk of type 2 diabetes in people with depression.   Diabetes Res Clin Pract. 2008;79(1):61-67. doi:10.1016/j.diabres.2007.07.009 PubMedGoogle ScholarCrossref
    19.
    Kivimäki  M, Batty  GD, Jokela  M,  et al.  Antidepressant medication use and risk of hyperglycemia and diabetes mellitus: a noncausal association?   Biol Psychiatry. 2011;70(10):978-984. doi:10.1016/j.biopsych.2011.07.008 PubMedGoogle ScholarCrossref
    20.
    Mezuk  B, Johnson-Lawrence  V, Lee  H,  et al.  Is ignorance bliss? depression, antidepressants, and the diagnosis of prediabetes and type 2 diabetes.   Health Psychol. 2013;32(3):254-263. doi:10.1037/a0029014 PubMedGoogle ScholarCrossref
    21.
    Azevedo Da Silva  M, Dugravot  A, Balkau  B,  et al; D.E.S.I.R. Study Group.  Antidepressant medication use and trajectories of fasting plasma glucose, glycated haemoglobin, β-cell function and insulin sensitivity: a 9-year longitudinal study of the D.E.S.I.R. cohort.   Int J Epidemiol. 2015;44(6):1927-1940. doi:10.1093/ije/dyv153 PubMedGoogle ScholarCrossref
    22.
    Pyykkönen  AJ, Räikkönen  K, Tuomi  T, Eriksson  JG, Groop  L, Isomaa  B.  Depressive symptoms, antidepressant medication use, and insulin resistance: the PPP-Botnia Study.   Diabetes Care. 2011;34(12):2545-2547. doi:10.2337/dc11-0107 PubMedGoogle ScholarCrossref
    23.
    Arterburn  D, Sofer  T, Boudreau  DM,  et al.  Long-term weight change after initiating second-generation antidepressants.   J Clin Med. 2016;5(4):E48. doi:10.3390/jcm5040048 PubMedGoogle Scholar
    24.
    Blumenthal  SR, Castro  VM, Clements  CC,  et al.  An electronic health records study of long-term weight gain following antidepressant use.   JAMA Psychiatry. 2014;71(8):889-896. doi:10.1001/jamapsychiatry.2014.414 PubMedGoogle ScholarCrossref
    25.
    Gafoor  R, Booth  HP, Gulliford  MC.  Antidepressant utilisation and incidence of weight gain during 10 years’ follow-up: population based cohort study.   BMJ. 2018;361:k1951. doi:10.1136/bmj.k1951 PubMedGoogle ScholarCrossref
    26.
    Burcu  M, Zito  JM, Safer  DJ,  et al.  Association of antidepressant medications with incident type 2 diabetes among Medicaid-insured youths.   JAMA Pediatr. 2017;171(12):1200-1207. doi:10.1001/jamapediatrics.2017.2896 PubMedGoogle ScholarCrossref
    27.
    Burcu  M, Zito  JM, Safer  DJ,  et al.  Concomitant use of atypical antipsychotics with other psychotropic medication classes and the risk of type 2 diabetes mellitus.   J Am Acad Child Adolesc Psychiatry. 2017;56(8):642-651. doi:10.1016/j.jaac.2017.04.004 PubMedGoogle ScholarCrossref
    28.
    Mansur  RB, McIntyre  RS, Brietzke  E.  Mental disorders, antidepressants, and type 2 diabetes.   JAMA Pediatr. 2018;172(7):703-704. doi:10.1001/jamapediatrics.2018.1005 PubMedGoogle ScholarCrossref
    29.
    Suglia  SF, Demmer  RT, Wahi  R, Keyes  KM, Koenen  KC.  Depressive symptoms during adolescence and young adulthood and the development of type 2 diabetes mellitus.   Am J Epidemiol. 2016;183(4):269-276. doi:10.1093/aje/kwv149 PubMedGoogle ScholarCrossref
    30.
    Shomaker  LB, Tanofsky-Kraff  M, Stern  EA,  et al.  Longitudinal study of depressive symptoms and progression of insulin resistance in youth at risk for adult obesity.   Diabetes Care. 2011;34(11):2458-2463. doi:10.2337/dc11-1131 PubMedGoogle ScholarCrossref
    31.
    Pinhas-Hamiel  O, Zeitler  P.  Acute and chronic complications of type 2 diabetes mellitus in children and adolescents.   Lancet. 2007;369(9575):1823-1831. doi:10.1016/S0140-6736(07)60821-6 PubMedGoogle ScholarCrossref
    32.
    Ivanovska  V, Rademaker  CM, van Dijk  L, Mantel-Teeuwisse  AK.  Pediatric drug formulations: a review of challenges and progress.   Pediatrics. 2014;134(2):361-372. doi:10.1542/peds.2013-3225 PubMedGoogle ScholarCrossref
    33.
    McMahon  AW, Dal Pan  G.  Assessing drug safety in children—the role of real-world data.   N Engl J Med. 2018;378(23):2155-2157. doi:10.1056/NEJMp1802197 PubMedGoogle ScholarCrossref
    34.
    ’t Jong  GW, Vulto  AG, de Hoog  M, Schimmel  KJ, Tibboel  D, van den Anker  JN.  Unapproved and off-label use of drugs in a children’s hospital.   N Engl J Med. 2000;343(15):1125. doi:10.1056/NEJM200010123431515 PubMedGoogle ScholarCrossref
    35.
    Frattarelli  DA, Galinkin  JL, Green  TP,  et al; American Academy of Pediatrics Committee on Drugs.  Off-label use of drugs in children.   Pediatrics. 2014;133(3):563-567. doi:10.1542/peds.2013-4060 PubMedGoogle ScholarCrossref
    36.
    Corny  J, Lebel  D, Bailey  B, Bussières  JF.  Unlicensed and off-label drug use in children before and after pediatric governmental initiatives.   J Pediatr Pharmacol Ther. 2015;20(4):316-328.PubMedGoogle Scholar
    37.
    Cheung  K, Teichert  M, Moll  HA, Stricker  BH, Visser  LE.  Filled prescriptions of age-related contraindicated drugs in children: a one-year nationwide cohort study in the Netherlands.   Int J Clin Pharm. 2018;40(5):1137-1143. doi:10.1007/s11096-018-0717-6 PubMedGoogle ScholarCrossref
    38.
    Hoon  D, Taylor  MT, Kapadia  P, Gerhard  T, Strom  BL, Horton  DB.  Trends in off-label drug use in ambulatory settings: 2006-2015.   Pediatrics. 2019;144(4):e20190896. doi:10.1542/peds.2019-0896 PubMedGoogle Scholar
    39.
    Lovegrove  MC, Weidle  NJ, Budnitz  DS.  Trends in emergency department visits for unsupervised pediatric medication exposures, 2004-2013.   Pediatrics. 2015;136(4):e821-e829. doi:10.1542/peds.2015-2092 PubMedGoogle ScholarCrossref
    40.
    Neubert  A, Dormann  H, Weiss  J,  et al.  The impact of unlicensed and off-label drug use on adverse drug reactions in paediatric patients.   Drug Saf. 2004;27(13):1059-1067. doi:10.2165/00002018-200427130-00006 PubMedGoogle ScholarCrossref
    41.
    Impicciatore  P, Choonara  I, Clarkson  A, Provasi  D, Pandolfini  C, Bonati  M.  Incidence of adverse drug reactions in paediatric in/out-patients: a systematic review and meta-analysis of prospective studies.   Br J Clin Pharmacol. 2001;52(1):77-83. doi:10.1046/j.0306-5251.2001.01407.x PubMedGoogle ScholarCrossref
    42.
    Lu  H, Rosenbaum  S.  Developmental pharmacokinetics in pediatric populations.   J Pediatr Pharmacol Ther. 2014;19(4):262-276.PubMedGoogle Scholar
    43.
    Sherman  RE, Anderson  SA, Dal Pan  GJ,  et al.  Real-world evidence—what is it and what can it tell us?   N Engl J Med. 2016;375(23):2293-2297. doi:10.1056/NEJMsb1609216 PubMedGoogle ScholarCrossref
    44.
    Osokogu  OU, Dukanovic  J, Ferrajolo  C,  et al.  Pharmacoepidemiological safety studies in children: a systematic review.   Pharmacoepidemiol Drug Saf. 2016;25(8):861-870. doi:10.1002/pds.4041 PubMedGoogle ScholarCrossref
    45.
    Schneeweiss  S, Avorn  J.  A review of uses of health care utilization databases for epidemiologic research on therapeutics.   J Clin Epidemiol. 2005;58(4):323-337. doi:10.1016/j.jclinepi.2004.10.012 PubMedGoogle ScholarCrossref
    46.
    Lund  JL, Richardson  DB, Stürmer  T.  The active comparator, new user study design in pharmacoepidemiology: historical foundations and contemporary application.   Curr Epidemiol Rep. 2015;2(4):221-228. doi:10.1007/s40471-015-0053-5 PubMedGoogle ScholarCrossref
    47.
    Patten  SB.  Bupropion confirmed not to cause weight gain in a real-world clinical population.   Evid Based Ment Health. 2015;18(2):57. doi:10.1136/eb-2014-101978 PubMedGoogle ScholarCrossref
    48.
    Cheung  AH, Zuckerbrot  RA, Jensen  PS, Laraque  D, Stein  REK; GLAD-PC STEERING GROUP.  Guidelines for Adolescent Depression in Primary Care (GLAD-PC): part II, treatment and ongoing management.   Pediatrics. 2018;141(3):e20174082. doi:10.1542/peds.2017-4082 PubMedGoogle Scholar
    49.
    Cipriani  A, Zhou  X, Del Giovane  C,  et al.  Comparative efficacy and tolerability of antidepressants for major depressive disorder in children and adolescents: a network meta-analysis.   Lancet. 2016;388(10047):881-890. doi:10.1016/S0140-6736(16)30385-3 PubMedGoogle ScholarCrossref
    50.
    Teltsch  DY, Fazeli Farsani  S, Swain  RS,  et al.  Development and validation of algorithms to identify newly diagnosed type 1 and type 2 diabetes in pediatric population using electronic medical records and claims data.   Pharmacoepidemiol Drug Saf. 2019;28(2):234-243. doi:10.1002/pds.4728 PubMedGoogle ScholarCrossref
    51.
    Desai  RJ, Rothman  KJ, Bateman  BT, Hernandez-Diaz  S, Huybrechts  KF.  A propensity-score-based fine stratification approach for confounding adjustment when exposure is infrequent.   Epidemiology. 2017;28(2):249-257. doi:10.1097/EDE.0000000000000595 PubMedGoogle ScholarCrossref
    52.
    Robins  J, Hernán  M, Fitzmaurice  G, Davidian  M, Verbeke  G, Molenberghs  G.  Longitudinal Data Analysis. Chapman and Hall/CRC Press; 2009:553-599.
    53.
    Hernán  MA, Lanoy  E, Costagliola  D, Robins  JM.  Comparison of dynamic treatment regimes via inverse probability weighting.   Basic Clin Pharmacol Toxicol. 2006;98(3):237-242. doi:10.1111/j.1742-7843.2006.pto_329.x PubMedGoogle ScholarCrossref
    54.
    Hernán  MA.  The hazards of hazard ratios.   Epidemiology. 2010;21(1):13-15. doi:10.1097/EDE.0b013e3181c1ea43 PubMedGoogle ScholarCrossref
    55.
    Murray  EJ, Hernán  MA.  Improved adherence adjustment in the Coronary Drug Project.   Trials. 2018;19(1):158. doi:10.1186/s13063-018-2519-5 PubMedGoogle ScholarCrossref
    56.
    Bobo  WV, Cooper  WO, Stein  CM,  et al.  Antipsychotics and the risk of type 2 diabetes mellitus in children and youth.   JAMA Psychiatry. 2013;70(10):1067-1075. doi:10.1001/jamapsychiatry.2013.2053 PubMedGoogle ScholarCrossref
    57.
    Galling  B, Roldán  A, Nielsen  RE,  et al.  Type 2 diabetes mellitus in youth exposed to antipsychotics: a systematic review and meta-analysis.   JAMA Psychiatry. 2016;73(3):247-259. doi:10.1001/jamapsychiatry.2015.2923 PubMedGoogle ScholarCrossref
    58.
    American Academy of Pediatrics Subcommittee on Attention-Deficit/Hyperactivity Disorder and Committee on Quality Improvement.  Clinical practice guideline: treatment of the school-aged child with attention-deficit/hyperactivity disorder.   Pediatrics. 2001;108(4):1033-1044. doi:10.1542/peds.108.4.1033 PubMedGoogle ScholarCrossref
    59.
    Cole  SR, Hernán  MA.  Constructing inverse probability weights for marginal structural models.   Am J Epidemiol. 2008;168(6):656-664. doi:10.1093/aje/kwn164 PubMedGoogle ScholarCrossref
    60.
    Schneeweiss  S, Rassen  JA, Glynn  RJ, Avorn  J, Mogun  H, Brookhart  MA.  High-dimensional propensity score adjustment in studies of treatment effects using health care claims data.   Epidemiology. 2009;20(4):512-522. doi:10.1097/EDE.0b013e3181a663cc PubMedGoogle ScholarCrossref
    61.
    Lipsitch  M, Tchetgen Tchetgen  E, Cohen  T.  Negative controls: a tool for detecting confounding and bias in observational studies.   Epidemiology. 2010;21(3):383-388. doi:10.1097/EDE.0b013e3181d61eeb PubMedGoogle ScholarCrossref
    62.
    Marmorstein  NR, Iacono  WG, Legrand  L.  Obesity and depression in adolescence and beyond: reciprocal risks.   Int J Obes (Lond). 2014;38(7):906-911. doi:10.1038/ijo.2014.19 PubMedGoogle ScholarCrossref
    63.
    Abbasi  A, Juszczyk  D, van Jaarsveld  CHM, Gulliford  MC.  Body mass index and incident type 1 and type 2 diabetes in children and young adults: a retrospective cohort study.   J Endocr Soc. 2017;1(5):524-537. doi:10.1210/js.2017-00044 PubMedGoogle ScholarCrossref
    64.
    Austin  PC.  Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.   Stat Med. 2009;28(25):3083-3107. doi:10.1002/sim.3697 PubMedGoogle ScholarCrossref
    65.
    Deuschle  M.  Effects of antidepressants on glucose metabolism and diabetes mellitus type 2 in adults.   Curr Opin Psychiatry. 2013;26(1):60-65. doi:10.1097/YCO.0b013e32835a4206 PubMedGoogle ScholarCrossref
    66.
    McIntyre  RS, Soczynska  JK, Konarski  JZ, Kennedy  SH.  The effect of antidepressants on glucose homeostasis and insulin sensitivity: synthesis and mechanisms.   Expert Opin Drug Saf. 2006;5(1):157-168. doi:10.1517/14740338.5.1.157 PubMedGoogle ScholarCrossref
    67.
    Dabelea  D, Mayer-Davis  EJ, Saydah  S,  et al; SEARCH for Diabetes in Youth Study.  Prevalence of type 1 and type 2 diabetes among children and adolescents from 2001 to 2009.   JAMA. 2014;311(17):1778-1786. doi:10.1001/jama.2014.3201 PubMedGoogle ScholarCrossref
    68.
    Gaskin  DJ, Thorpe  RJ  Jr, McGinty  EE,  et al.  Disparities in diabetes: the nexus of race, poverty, and place.   Am J Public Health. 2014;104(11):2147-2155. doi:10.2105/AJPH.2013.301420 PubMedGoogle ScholarCrossref
    69.
    Scott  RA, Langenberg  C, Sharp  SJ,  et al; InterAct Consortium.  The link between family history and risk of type 2 diabetes is not explained by anthropometric, lifestyle or genetic risk factors: the EPIC-InterAct study.   Diabetologia. 2013;56(1):60-69. doi:10.1007/s00125-012-2715-x PubMedGoogle ScholarCrossref
    70.
    Dabelea  D, Mayer-Davis  EJ, Lamichhane  AP,  et al.  Association of intrauterine exposure to maternal diabetes and obesity with type 2 diabetes in youth: the SEARCH Case-Control Study.   Diabetes Care. 2008;31(7):1422-1426. doi:10.2337/dc07-2417 PubMedGoogle ScholarCrossref
    71.
    Ferguson  JM.  SSRI antidepressant medications: adverse effects and tolerability.   Prim Care Companion J Clin Psychiatry. 2001;3(1):22-27. doi:10.4088/PCC.v03n0105 PubMedGoogle ScholarCrossref
    72.
    Shatin  D, Levin  R, Ireys  HT, Haller  V.  Health care utilization by children with chronic illnesses: a comparison of Medicaid and employer-insured managed care.   Pediatrics. 1998;102(4):E44. doi:10.1542/peds.102.4.e44 PubMedGoogle Scholar
    73.
    Kelsey  JL, Whittemore  AS, Evans  AS, Thompson  WD.  Methods in Observational Epidemiology. Monographs in Epidemiology and Biostatistics; 1996.
    74.
    Huitfeldt  A, Hernan  MA, Kalager  M, Robins  JM.  Comparative effectiveness research using observational data: active comparators to emulate target trials with inactive comparators.   EGEMS (Wash DC). 2016;4(1):1234. doi:10.13063/2327-9214.1234 PubMedGoogle Scholar
    75.
    Boudreau  DM, Arterburn  D, Bogart  A,  et al.  Influence of body mass index on the choice of therapy for depression and follow-up care.   Obesity (Silver Spring). 2013;21(3):E303-E313. doi:10.1002/oby.20048 PubMedGoogle ScholarCrossref
    ×